Autoguider Release note Summary

The Autoguider system has been through many releases in the near past. This document summarises the various releases to date.

Major Releases

Version 1 was the original system delivered to LaPalma when the telescope was being commissioned. No details are known.
Version 2 is undocumented, I suspect that it included support for the UES instrument as a separate program.
Version 3, dating from 1993, was the first system to be consistently developed, with work on the Display drivers and all other areas of the system to improve reliability.
Version 4, was released in 1994 and included a port to ImageForth Version 4.2 and operation with the OmniComp graphics card. Operation over the Utility network became reliable and all applications / configurations became available without haveing to reload software. Focal stations for Prime Focus, AutoFib and Ghril were added. The AutoFib configuration involved most change to the system and much more rigorous testing than had been done until then.

Detailed release notes

This section gives a brief summary of each release.

4.0.2c

operational system prior to Af2 commissioning visit in September
requires CCD firmware 2.2 with non SCSI disk card.
Used at CASS, UES ( old A&G ), Prime and Ghril (once)
Uses D-task V3-1

4.0.2d

Version of operational code modified slightly to run with the upgraded CCDc firmware V3.1 Never used other that brief tests to confirm basic operation. This was produced as a 1st level fallback position.

4.1.1

First preoperational system used for sky tests. lots of differences

• new sky estimation with histograms
• new centroid code extracts star from peak to 1 sigma point.
• Af2 routines rewritten from 4.0.x
• uses new D-task 4-0B7 with lots more commands, and some parameter changes.
• lab tests performed for af2 and cass operations

4.1.1b

mod to dark current file save routines to allow them to save the frame to the current disk, not just the hard disk. used on sky 13/9/94 02:00 to 05:30

4.1.2

mod to TCS packet to include readout time in delay estimate to stop TCS timeouts.

• guiding with multiple windows now working, window can track.
• default guide window size=30. used on sky 13/9/94 05:30 - 07:15

4.1.3

CASS field acquisition area enlarged to almost full frame scale factors to stop centroid sums overflowing now taken account of in magnitude calcs.

• SEND-BLOCKS included for transfer to sparc via serial line.
• AG-BACKUP added for easy backup to floptical, makes floptical bootable without any hassle. D-task 4-0B7 CCDc 3.1 Used on sky 14/9/94 onwards as operational system at Cass.

4.1.4

UES acquisition and guiding code simplified to act as Cass system with Field Select guide method.

• 0's in histogram ignores in sky estimation
• window size changing with 2 windows active now works
• multitasking real & int display code fixed which should affect stats.
• dark-patterns algorithm uses DCA buffers, not im stack
• guid plots with scale on bottom axis
• fiducial point displayed with box.

4.1.5

• plot routine > 156 samples caused guide process to fall over - now fixed. Plot loop recoded to stop double plotting
• slit centroid now centroids whole window, slit guiding now works.
• time between VAx packets now calculated better.

4.1.6

guiding routines do not draw cursor when CCDCUR is running to allow CCDCU to work with guide loop on. Used on t/s 20/9/94 20:30 --> The following versions date from my second AF2 commissioning trip in October.

4.1.7

• UES window within readout window exctraction round peak re-written not to cause 0 size windows ( should fix slit guiding )
• ccd readout time routine recoded to work better.
• Elapsed time between TCS packets now calculated so that timeouts can be detected when debugging.
• Bug in TCS packet code found & removed. ( caused guide hang )
• stats routine uses display task to print variables on screen
• tracking window calls correct routine when 2 windows active
• CEN cmd hanging when system idle ( no CTA no GUI ) bug fixed.
• 0 sample value in histogram restored.
• SEL marks star with big box - recoded.
• peak centroid routine ( used by UES ) recoded so that if the star is near the edge of the window, the box extracted is shifted towards the centre of the window, not truncated by the window edge. This caused 0 size boxes at UES
• MEdian estimate recoded to cope with images full of 0's
• image scaling to avoid centroid overflow moved to after the median / sky / magnitude estimate routines to improve the accuracy of the sky estimate. This was poor with bright stars since sky was scaled down to 0,1,2 pixel counts, and accuracy lost.

4.1.8

• SAW : test for 0 size acquisition window in SAW
• WMO : calls correct re-config routine with 2 windows active.
• All full frame readouts calculate & update ccd bias value.

4.1.9

Bisto! found bug causing v. occasional hang when doing re-configure of windows. CCD CRM-ON & CRM-OFF code called by routines which check that loop is off before doing something in the CCD controller which is illegal wile the loop is running. Moved CHK-CRM up in hierarchy to call CRM-OFF not the lower level CCD-CRM. This should improve reliability.

• GLP and INT now send an extra packet to the TCS to inform of new delay time when conditions change in AG. ( uses last star centroid value. )

4.1.10

• CWN while CEN running caused re-config and used new CWN window - which gave an offset in the returned CEN data. Local copy of CWN windwo used during CEN operation.
• Same magnitude for both CEN stars reported: found bug so that individual magnitudes stored and reported.

4.1.11

CWN now accepts any two points in fibre space, not restricted to x1 < x2 and y1 < y2

4.1.12

CEN aborts correctly if the GUI or CTA loops used to get the CCD data abort themselves.
CTA mechanism error cleared by CEN command if CTA not running when CEN starts - stops CEN aborting from previous CTA error.

4.1.13

CWN tests for 0 0 0 0 or 0 size window - recoded.
CWN tests for both points in fibre area
CWN tests for window in scanned area of CCD as well as within fibre.
WMO tests for window in scanned area of CCD. .. ..

4.1.14

Black frame on TV bug. Gain and offset to convert full frame dark current redaout pattern into a suitable one for the particular windowed readout in use are calculated. The offset is calculated from the average pixel value in the bottom row of the window area extracted from the dark frame. -ve pixels were treated as unsigned large numbers => large offset => removed from image as DC level then clipped to 0. Ouch! Row average routine SET-RO-SCALES modified to do sign extension and treat data as 16 bit signed. ALso changes sense of offset when added to image data. THIS BUG ALSO FIXED IN V4.1.3 => 4.1.3a 4.1.14 tested extensively in electronics lab - a few bugs found. worked with acquire i-task ok.

4.1.15

• magnitude calc now copes with 0 int time.
• FLD clears sel mech error to stop complaining about previous select failure when selecting the brightest star.
• SEL marker clearing re-coded ( still doesnt work properly )
• GLP INT only send extra packet to TCS if delay increases.
• CWN has no effect until next CEN command. - recoded slightly
• GDM reports error if CEN running.

4.1.16

10/10/94 GSX-BOX stored local copy of position for erase; GMOVE' GLINE' GBOX' now limit to graphics page size.

4.1.17

14/10/94 window & label drawing now done via entity table, all you do is update the table flags. Refresh of visible windows now available.

4.1.18

4/10/94 bug in gsx-box, not being drawn. UES WMO bug in win-origin

4.1.19

15/10/94
network : ?CBUSY does not set bit 0 if busy.
ccd iface: ccd-init called ccd-crm not crm-off general : ag-saw redraws acq win & label

4.1.20

16/10/94 implemented separate cursors for fixed&mobile CRC. Removed readout err flag test on normal track loop exit. Added chk-crm to ccd-cwi. Added crm-off to fatal end of track-loop

4.1.21

24/10/94 added #ghril to app>port
Made seeing disp = -.-- for 0.00 to flag invalid seeing value. Field slit box = 50 x 100 for search. Slit cent now uses 2sig mask in y and star-cent in x. Slit cent makes median=2sig point for d-cog call.
Smaller prompt on engineering terminal, now shows Version and time, for example A/G V4.1.21 15:47:23 >

4.1.22

Added d-smooth to centroid after sky estimate, not before.

4.1.23

Overflow bug in d-cog fixed. Cursor & window not drawn at bott left of box

4.1.24

non cambridge ngi load uncommented block 1061
Guide loop abort will now display dici error status

4.1.25

Unsigned version of d-smooth now used d-usmooth. Dark current averages 5 frames to stop overflow with large ccd bias value.

4.1.26

SLIT-CENT actually calls SLITCENT by mistake. Values for low signal and suspension thresholds empirically set: centroid-thresh = 12.00 for slit, 3.00 otherwise fibre border = 2 pix DISK->CHT now does separate reads per disk block 4.1.26: Latest centroid algorithm with local sky estimation. UES slit has simple centroid in the Y direction, without star extraction.

4.1.27

4.1.27: System beeps when going into suspended mode.

4.1.28

4.1.28: old v4.0.2d centroid added for experiment

4.1.29

Same functionality as V4.1.28 but coded more elegantly & tested further.

4.1.30

?NET operation adjusted, AUT1 address restored, Continuous window update when at left hand edge of fibre corrected.

4.2

Development version before txfer to LPO for UES slit guiding testing Feb 94.

4.3

Improved UES slit performance and sensitivity. following campaign on LPO. UES operational version.

4.4

Operational version for use with AF2 instrument. Should be compatible at UES, CASS GHRIL and PRIME focii.